Assessment of Insect Communities with Metabarcoding

Abstract Metabarcoding is widely used for molecular identification of organisms. Due to its efficiency for en masse characterization of samples, the technique is useful for insect biodiversity surveys. Although metabarcoding has been used for nearly two decades, there is still a need for optimized insect sample processing strategies. The goal of this study was to establish best practices for molecular characterization of bulk insect collections. We sampled insect diversity using light traps in lowland dipterocarp forest of Tengku Hassanal Wildlife Reserve in Pahang, Malaysia. Each light trap sample was homogenized and repeatedly subsampled to identify the number of subsamples required to detect total estimated insect OTU diversity in a light trap. Insect OTU diversity from 72 subsamples (12 subsamples from each of six light traps) was characterized by sequencing part of the mitochondrial Cytochrome Oxidase I (COI) gene. We found that four and eight 100 µL subsamples were sufficient to detect at least 90% and 95%, respectively, of insect OTU diversity collected in each light trap sample, regardless of the range of variation in preserved biomass (137.9 ̶ 445.4 g). We also built a global database from the Barcode of Life Database (BOLD) repository for our target sequencing region (313bp). We found that standardized bitscores of tblastx were significantly higher than blastn (p < 0.001); however, the percent identity distributions were not statistically different, with both around 92–93%. In addition to relatively low average match identities, we also found poor taxonomic concordance between blastn and tblastx, especially at lower taxonomic levels, and suggest the advantage of metabarcoding should be leveraged by focusing on phylogenetic diversity instead of taxonomy wherever possible.